Pairing-free Group-level Knowledge Distillation for Robust Gastrointestinal Lesion Classification in White-Light Endoscopy

2026-01-14

AI Summary

This work addresses the challenge of cross-modal knowledge transfer from narrow-band imaging (NBI) to white-light endoscopy (WLI) in the absence of image-level paired data. To this end, the authors propose PaGKD, a novel framework that, for the first time, enables effective transfer of NBI diagnostic knowledge to WLI models without requiring pixel-wise or image-level correspondences. The approach leverages group-level prototype distillation (GKD-Pro) and dense distillation (GKD-Den), integrated with shared lesion-aware queries and activation-derived relational graphs, to align local structures while preserving global semantic consistency. Evaluated on four clinical datasets, the method achieves consistent performance gains, improving AUC by 3.3%, 1.1%, 2.8%, and 3.2% respectively, significantly outperforming existing state-of-the-art approaches.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Multimodal Learning
• Natural Language Processing: Language Grounding & Multi-modal NLP

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

White-Light Imaging (WLI) is the standard for endoscopic cancer screening, but Narrow-Band Imaging (NBI) offers superior diagnostic details. A key challenge is transferring knowledge from NBI to enhance WLI-only models, yet existing methods are critically hampered by their reliance on paired NBI-WLI images of the same lesion, a costly and often impractical requirement that leaves vast amounts of clinical data untapped. In this paper, we break this paradigm by introducing PaGKD, a novel Pairing-free Group-level Knowledge Distillation framework that that enables effective cross-modal learning using unpaired WLI and NBI data. Instead of forcing alignment between individual, often semantically mismatched image instances, PaGKD operates at the group level to distill more complete and compatible knowledge across modalities. Central to PaGKD are two complementary modules: (1) Group-level Prototype Distillation (GKD-Pro) distills compact group representations by extracting modality-invariant semantic prototypes via shared lesion-aware queries; (2) Group-level Dense Distillation (GKD-Den) performs dense cross-modal alignment by guiding group-aware attention with activation-derived relation maps. Together, these modules enforce global semantic consistency and local structural coherence without requiring image-level correspondence. Extensive experiments on four clinical datasets demonstrate that PaGKD consistently and significantly outperforms state-of-the-art methods, achieving relative AUC improvements of 3.3%, 1.1%, 2.8%, and 3.2%, respectively, establishing a new direction for cross-modal learning from unpaired data.

Problem

• knowledge distillation
• unpaired data
• cross-modal learning
• gastrointestinal lesion classification
• white-light endoscopy

Innovation

• pairing-free knowledge distillation
• group-level distillation
• cross-modal learning
• unpaired medical images
• endoscopic lesion classification